Sandia National Labs FY20 LDRD Annual Report


Efficient real-time cognition at the point of sensing. Sensors operating on the principles of “compressive sensing” (CS) can accurately represent a signal with dramatically fewer data points than traditional sensors. However, translating this data into a form humans can interpret requires a computationally-intensive reconstruction algorithm. Traditionally, machine learning algorithms have been applied to CS systems after the reconstruction step. Sandia researchers asked the question, “If a machine is performing the analysis, why put the data in a format a human can understand?” They successfully demonstrated that machine learning techniques can be employed on CS data in their native domain without the reconstruction. This reduces power and computational requirements and enhances algorithm performance and speed. This new sensing paradigm will enable faster, more efficient automated systems for a wide variety of national security missions. (PI: Eric Shields)

Data collected from sensors aboard HOT SHOT sounding rockets is allowing researchers to improve computer- and ground-based simulations of flight vibrations. Ralph Lied-Lopez helped to study the amount of vibration that mechanical objects endure in flight, including the so-called “wedding cake” hardware, a mock rocket component. (Photo by Norman Johnson)



Made with FlippingBook - professional solution for displaying marketing and sales documents online